https://d2ng68cf36kxsk.cloudfront.net/store/public/post/1228/post_images/Slide402.JPG">

The man they interviewed in the research lab said that within five years, “We would be able to do … about one to two million words an hour, and this would be quite an adequate speed to cope with the whole output of the Soviet Union in just a few hours of computer time a week.”⁸⁰

Yehoshua Bar-Hillel organized the first conference on machine translation at MIT in September 1952, and military sponsorship of the research was in the air. In his introductory chapter to the published volume of papers, Warren Weaver drew connections between wartime cryptography and machine translation.⁸¹ Weaver was famous for his work with Claude Shannon^§MS1.3.2b in The Mathematical Theory of Communication.⁸² Papers addressing specific language translation issues were mostly about translation from Russian.⁸³ The US Government and military poured many millions of dollars into machine translation in the 1950s. The Soviets responded with substantial investment for their researchers.⁸⁴

Then came doubts. Bar-Hillel returned to Israel, where he took up a position at Hebrew University. His successor in the MIT Research Laboratory of Electronics, Victor Yngve, started off equally convinced that Chomsky’s approach was worth pursuing. But try as they may, he and others in the lab couldn’t make it work. Chomsky’s program was “based on logical arguments and a priori assumptions about requirements of an abstract grammar, [and] ignored the fact that sentences are produced by people using a finite processor [with] … limited temporary memory.” And although Chomsky’s grammar was “meant to be generative in the mathematical sense, [it] did not actually construct sentences one at a time to order in real time.”⁸⁵

Then Bar-Hillel expressed his doubts about the whole project. Writing now in 1959, “A few years ago, I proposed what I called a quasi-arithmetical notation for syntactic description … whose employment should allow, after some refinements, for a mechanical determination of the constituent structure of any given sentence. I am now quite convinced that this hope will not come true.”⁸⁶

In a “Demonstration of the Nonfeasibility of Fully Automatic High Quality Translation,” Bar-Hillel offered the sentence, “The box was in the pen.” The pen in this case was not the writing instrument, but the enclosure in which small children play, the playpen. “I now claim that no existing or imaginable program will enable an electronic computer to determine that the word ‘pen’” has this meaning. However, a human with knowledge of the relative sizes of different kinds of pens, will. “This knowledge is not at the disposal of the electronic computer.”⁸⁷ Here, Bar-Hillel returns to the empirical, observable, semantic world that he (in a former life) and Carnap had all-but abandoned in their quest to develop a mechanical analytics, as Chomsky also did in his generative theory.

Finally came abandonment of the project by the US military-scientific establishment. In 1965, the Department of Defense, the Central Intelligence Agency, and the National Science Foundation formed a committee to review progress “on research and development in the general field of mechanical translation of foreign languages.” An awful lot of money had been spent, the report concluded, and even with human post-editing the result was “machine translations [that] are slow, expensive, of poor graphic arts quality, and not very good translations.” Chomsky’s work had not been successfully applied to computers and, said the committee, “[t]here can be no doubt that the disappointingly slender computer results realized on the basis of [linguistic] theory must have been important in shaking at least some inquisitive linguists out of their contentment.”

Besides, there was plenty of capacity for human translation of Russian. “There is no emergency in the field of translation. The problem is not to meet some nonexistent need through nonexistent machine translation.”⁸⁸

---

⁸⁰ www.youtube.com/watch?v=K-HfpsHPmvw

⁸¹ Warren Weaver, 1949 [1955], “Translation,” pp. 15–23 in Machine Translation of Languages, edited by W.N. Locke and A.D. Booth, Cambridge, MA: The Technology Press of MIT, pp. 16–17, 22–23.

⁸² Claude E. Shannon and Warren Weaver, 1949, The Mathematical Theory of Communication, Urbana: University of Illinois Press.

⁸³ William N. Locke and A. Donald Booth, eds. 1955, The Machine Translation of Languages, Cambridge, MA: The Technology Press of MIT.

⁸⁴ W. John Hutchins, ed. 2000a, Early Years in Machine Translation: Memoirs and Biographies of Pioneers, Amsterdam: John Benjamins, pp. 197ff.

⁸⁵ Victor H. Yngve, 2000, “Early Research at M.I.T.,” pp. 39–72 in Early Years in Machine Translation: Memoirs and Biographies of Pioneers, edited by W.J. Hutchins, Amsterdam: John Benjamins, pp. 64–65.

⁸⁶ Yehoshua Bar-Hillel, 1959, “Report of the State of Machine Translation in the United States and Great Britain,” Jerusalem: Hebrew University, Appendix II, p. 1.

⁸⁷ Yehoshua Bar-Hillel, 1960, “The Present Status of Automatic Translation of Languages,” Advances in Computers 1:91–163, pp. 158–60.

⁸⁸ Automatic Language Processing Advisory Committee, 1966, “Language and Machines: Computers in Translation and Linguistics,” Washington, DC: National Research Council, pp. 44, 123, 16.

---

Kalantzis, Mary, and Bill Cope. 2020. Adding Sense: Context and Interest in a Grammar of Multimodal Meaning. Cambridge UK: Cambridge University Press. pp. 213–215. https://doi.org/10.1017/9781108862059

The Automatic Language Processing Advisory Committee §AS2.2c